EU AI new law fines ChatGPT and Claude! Pirated training and infringement can be fined up to 7% of revenue

avatar
ABMedia
07-11
This article is machine translated
Show original

The European Commission published the General-Purpose AI Code of Practice on 7/11, serving as a transitional mechanism to help companies comply with the AI Act. The content focuses on protecting copyrights and increasing transparency of large AI models. It will be applicable to large generative AI models like ChatGPT and Claude as early as August, with potential fines up to 7% of revenue for violations.

Prohibit AI Training with Pirated Data to Protect Creators' Rights

The European Commission stated that the practice guidelines require developers not to use pirated content for AI training. If creators explicitly state that they do not want their works or data to be used for AI training, developers must respect and exclude such content.

Additionally, if AI-generated content raises copyright infringement suspicions, companies must have a mechanism to address and correct such issues.

Developers Must Provide Complete AI Documentation to Implement Transparency Principles

The Commission added that AI model developers must provide the latest model usage documentation to regulatory bodies and third-party integrators, specifically disclosing the operation and limitations of AI to implement the "transparency principle".

This regulation primarily targets "General Purpose AI" models, such as OpenAI's ChatGPT and Anthropic's Claude, expected to be implemented in August.

Guidelines Are Voluntary, with Potential Fines of Up to 7% of Annual Revenue

These guidelines are "voluntary" in nature, with companies able to choose whether to sign. The purpose is to help businesses prepare for the upcoming AI Act. The Commission emphasizes that companies signing the guidelines will receive more legal protection and reduce future legal risks.

Violations of the AI Act can result in fines up to 7% of the company's annual revenue. For developers of advanced AI models, the maximum fine is 3% of annual revenue. Originally planned for release in May, the Commission delayed the guidelines until July, with no current plans to postpone the law's implementation.

The guidelines were drafted by the European Commission, involving a working group of representatives from AI laboratories, tech companies, academic institutions, and digital rights organizations.

European and American Tech Companies Express Opposition

However, the guidelines immediately sparked backlash from tech companies like Meta and Alphabet, criticizing the initial draft for exceeding the original AI Act scope and adding extra burden on businesses. Local European companies have also voiced protests.

Companies like ASML, Airbus, and Mistral AI jointly called on the EU in early July to delay the AI Act implementation by two years, advocating for a more innovation-friendly regulatory approach.

Ongoing Debate on Balancing Creators' Rights and Technological Innovation

Previously, legendary Studio Ghibli director Hayao Miyazaki stated:

"I absolutely do not want to incorporate AI into my work. It is an insult to life itself. I feel humans are approaching the end of an era and losing confidence in themselves."

As AI models heavily rely on web-crawled data and potentially use creators' works without authorization, this not only risks copyright infringement but also makes many artists feel their creative value is diluted and emotions replaced. The EU's pioneering AI regulations and practice guidelines, despite ongoing controversies, undoubtedly attempt to clarify the boundaries of "data authorization", "technological responsibility", and "respecting creation". How to balance protecting creators, encouraging innovation, and preventing abuse will remain a long-term challenge for legal systems and industries worldwide.

Risk Warning

Cryptocurrency investment carries high risks, with prices potentially experiencing extreme volatility. You may lose your entire principal. Please carefully assess the risks.

Coinbase announced a collaboration with generative AI model Perplexity, with plans to integrate market data in stages. Future developments may include automated ordering, asset management, and even assistance in selecting cryptocurrencies and identifying buying opportunities.

Toggle

Coinbase Integrates Real-Time Information with Perplexity

On July 10 Eastern Time, Coinbase CEO Brian Armstrong announced a two-phase collaboration with Perplexity on Twitter (X). In the first phase, Perplexity will integrate Coinbase's real-time market information, with the "COIN50 Index" prices and related information already embedded in Perplexity's newly launched Comet browser. Here are the author's test screenshots:

  1. Left Image: When searching for crypto market data on Perplexity, AI will directly cite information sources provided by Coinbase.
  2. Right Image: Switching to the "Finance" feature on Perplexity will display "Crypto", and after clicking, the Coin50 Index will be prioritized at the top of the page.

Armstrong added that the next stage will arrive "very soon", at which point users asking about cryptocurrency information on Perplexity will be directly linked to Coinbase's market information, making the answers more real-time and accurate.

The Perplexity team also shared an interesting statistic: the number of users searching for cryptocurrencies is now equal to those searching for stocks, indicating that cryptocurrencies are accelerating towards mainstream adoption.

(Note: COIN50 is a representative index of 50 cryptocurrencies launched by Coinbase.)

Perplexity Valued at $14 Billion, Reportedly Seeking to Acquire TikTok's US Operations

Perplexity was founded in August 2022 by former OpenAI researcher Aravind Srinivas, focusing on real-time citation of internet data to answer questions, thus quickly rising in the AI search field.

Although the official user count has not been disclosed, the founder revealed in a January interview with Bloomberg that the platform has over 10 million users, with investors including NVIDIA, Databricks, and NEA. According to reports, Perplexity conducted a new round of $500 million funding in May this year, with the company's valuation expected to reach $14 billion after this round.

Just a day before announcing the collaboration with Coinbase, Perplexity had just launched Comet, an AI web browser challenging Google's search market. In January, Perplexity was also reportedly seeking to acquire TikTok's US operations.

AI Agents Will Definitely Become Users' "Exclusive Trading Managers"

Armstrong subsequently revealed that users will be able to use the following features through Perplexity:

  • Real-time tracking of market dynamics
  • Screening potential trading opportunities
  • Analyzing price trends of specific tokens

In other words, AI will become users' "exclusive trading manager" in the crypto market. In fact, as early as August last year, Armstrong had already said:

"Large language models (LLMs) should have crypto wallets. Let AI agents complete work and participate in transactions on behalf of users."

And a few weeks later, he personally supervised the first crypto trade completed entirely by AI.

(AI Search Rising Star Perplexity Valued at $18 Billion! SoftBank, NVIDIA, and Amazon Invest, Targeting Google and OpenAI)

Risk Warning

Cryptocurrency investment carries high risks, and prices may fluctuate dramatically. You may lose all your principal. Please carefully assess the risks.

Source
Disclaimer: The content above is only the author's opinion which does not represent any position of Followin, and is not intended as, and shall not be understood or construed as, investment advice from Followin.
Like
Add to Favorites
Comments